A study was conducted to investigate the effects of interaction analysis feedback on the verbal behavior of student teachers. Forty randomly selected senior elementary education students at the University of South Florida were trained in interaction analysis, and each was observed for eight weekly 15-minute periods during student teaching. Data, collected from systematic observations by four trained observers using a 13-category modification of the Flanders System of Interaction Analysis, were submitted to an IBM 1410 computer which produced matrices and information sheets weekly for the 40 students. Only the 20 designated as the "feedback group" (experimental) received the weekly matrices and information sheets. Data were analyzed by means of a Lindquist Type I analysis of variance, and F-tests for group means difference were computed for each of the 24 selected dependent variables. Findings included these: The experimental group used significantly less teacher-initiated talk and significantly more student-initiated talk than did the control group. They also used more praise, more extended use of student ideas, and less corrective feedback. They accepted and clarified student ideas more, criticized students less, asked more questions, and gave fewer directions. (Implications for preservice and inservice programs are discussed.) (JS)